The field of predictive analytics developed largely in the 1940s with governments first adopting computational models in various parts of their departments. The method of predicting future outcomes and events involves predictive models, machine learning and data mining. A notable example of predictive analytics is credit scoring, which has been used by financial service companies to examine the likelihood of individuals making debt repayments in time (Nyce and Cpcu, 2007).
The use of big data in business is nothing new, it has been used for decades as a means of getting better business performance. According to Li (2015), the use of big data has developed from primarily for reporting, to analysis then monitoring to prediction. The latter stage is where we are with predictive analytics usage in the EHS field. The global predictive analytics market is set for accelerated growth over the next decade (Johnson, 2019).
Whilst the use of predictive analytics in industries like finance and insurance is common, its use by EHS professionals is a somewhat recent phenomenon. With that being said, there is a large amount of data and information available at worksites that can be input into predictive models. Incident reports, sickness records, risk assessments and environmental readings can are all forms of historic data that can be used to predict future performance (PwC, 2014).
The use of predictive analytics in health and safety
Preventing workplace incidents has to be at the top of the list of priorities for any EHS professional, however many of the current data analysis in the world of health and safety focus on the leading indicators of incidents. The work of Mills, Turner and Pettinger (2017) suggests that moving towards the use of predictive indicators with the help of predictive analytics will better assist organisations in working out if and when safety incidents may occur.
Much of the current thought around the use of predictive analytics in EHS is on the need for a uniform process and standardisation, this led to the creation of Cross-Industry Standard Process. This methodology is also used in the wider field of predictive analytics.
There are some recorded case studies where the use of predictive analytics has helped to significantly reduce the threat of incidents. An ‘innovative tool for prediction of process upsets and hazard events’ developed by an oil company has the potential to warn of potentially dangerous events ‘hours in advance’ of issues (Cadei, et. al., 2018). The tool uses historical data, real-time statistics amongst other variables to produce reliable predictions on future events.
Future potential capabilities of predictive analytics
Whilst the use of statistical models analysing data in EHS is present many companies are currently operating towards the lower level of analytical maturity, in the operating and reporting functions. Whilst these are important moving towards the more advanced levels could significantly increase the usefulness of the data.
Predictive analytics is only useful if the sources of the data input into the necessary models are relevant. With the increase in technological development in EHS through the use of wearables, artificial intelligence and increased remote control, comes an increase in the scale and accuracy of potential data. This will allow organisations to better tailor their work practices to follow a statistically safe model. The data will then be able to indicate if a particular venture has been worthwhile, both in terms of saving lives and maintaining profitability.
Steps to consider before you implement predictive analytics
If a company decides to start using predictive analytics in the way it manages the workforce there are a number of steps that need to be followed to ensure a seamless transition.
1. Standardised data collection and storage across the organisation
Depending on the size of your organisation, the storage of vast amounts of data may require a certain level of IT infrastructure in place before the process can begin. The first step is to consult the IT department and ensure all the necessary hard and software is in place. The next step is to ensure everyone is clear on data collection.
2. Keep the process for collecting data as simple as possible at the uptake
At the end of the day if the statistics used in predictive models are not reliable the end forecasts will not be reliable. Make sure your organisation standardises the collection process. This will reduce the total number of errors and increase the reliability of the data.
3. Have an actual strategy for implementing the use of predictive analytics
One of the big failures many organisations encounter when using predictive analytics is where to go once the data has been collected and the predictions have been made. To overcome this make sure you put the business case forward for any recommended changes and ensure the changes are implemented throughout the organisation.
Predictive analytics is set to grow in multiple fields
As the tools and techniques used in predictive analytics become more sophisticated the data we get from the models will become more reliable. This will lead to an increase in the use of predictive analytics in fields like health and safety and education. Staying at the cutting edge of this statistical evaluation method will become increasingly crucial if businesses wish to stay competitive.
References
PWC (2014) Available here Nyce, C. and Cpcu, A., 2007. Predictive analytics white paper. American Institute for CPCU. Insurance Institute of America, pp.9-10.
Mills, T., Turner, M. and Pettinger, C., 2017. Advancing predictive indicators to prevent construction accidents. In Towards better Safety, Health, Wellbeing, and Life in Construction (pp. 459-466). Central University of Technology, Free State.
